{
  "best_metric": 0.9910973906517029,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 1.9184652278177459,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.009592326139088728,
      "grad_norm": 7.896941661834717,
      "learning_rate": 1e-05,
      "loss": 4.2978,
      "step": 1
    },
    {
      "epoch": 0.009592326139088728,
      "eval_loss": 1.4538829326629639,
      "eval_runtime": 13.7694,
      "eval_samples_per_second": 12.782,
      "eval_steps_per_second": 3.196,
      "step": 1
    },
    {
      "epoch": 0.019184652278177457,
      "grad_norm": 8.922088623046875,
      "learning_rate": 2e-05,
      "loss": 5.173,
      "step": 2
    },
    {
      "epoch": 0.02877697841726619,
      "grad_norm": 8.183208465576172,
      "learning_rate": 3e-05,
      "loss": 5.0288,
      "step": 3
    },
    {
      "epoch": 0.03836930455635491,
      "grad_norm": 6.656280517578125,
      "learning_rate": 4e-05,
      "loss": 4.9239,
      "step": 4
    },
    {
      "epoch": 0.047961630695443645,
      "grad_norm": 7.028189182281494,
      "learning_rate": 5e-05,
      "loss": 5.2642,
      "step": 5
    },
    {
      "epoch": 0.05755395683453238,
      "grad_norm": 6.835341930389404,
      "learning_rate": 6e-05,
      "loss": 4.8129,
      "step": 6
    },
    {
      "epoch": 0.0671462829736211,
      "grad_norm": 8.956515312194824,
      "learning_rate": 7e-05,
      "loss": 4.8395,
      "step": 7
    },
    {
      "epoch": 0.07673860911270983,
      "grad_norm": 7.3538126945495605,
      "learning_rate": 8e-05,
      "loss": 5.1888,
      "step": 8
    },
    {
      "epoch": 0.08633093525179857,
      "grad_norm": 7.477017879486084,
      "learning_rate": 9e-05,
      "loss": 5.2333,
      "step": 9
    },
    {
      "epoch": 0.09592326139088729,
      "grad_norm": 6.610411643981934,
      "learning_rate": 0.0001,
      "loss": 4.7623,
      "step": 10
    },
    {
      "epoch": 0.10551558752997602,
      "grad_norm": 6.699337959289551,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.8977,
      "step": 11
    },
    {
      "epoch": 0.11510791366906475,
      "grad_norm": 7.490584850311279,
      "learning_rate": 9.997266286704631e-05,
      "loss": 5.1014,
      "step": 12
    },
    {
      "epoch": 0.12470023980815348,
      "grad_norm": 6.455275535583496,
      "learning_rate": 9.993849845741524e-05,
      "loss": 4.8776,
      "step": 13
    },
    {
      "epoch": 0.1342925659472422,
      "grad_norm": 6.904367923736572,
      "learning_rate": 9.989068136093873e-05,
      "loss": 4.7319,
      "step": 14
    },
    {
      "epoch": 0.14388489208633093,
      "grad_norm": 7.001408100128174,
      "learning_rate": 9.98292246503335e-05,
      "loss": 4.8072,
      "step": 15
    },
    {
      "epoch": 0.15347721822541965,
      "grad_norm": 7.208923816680908,
      "learning_rate": 9.975414512725057e-05,
      "loss": 4.9104,
      "step": 16
    },
    {
      "epoch": 0.1630695443645084,
      "grad_norm": 7.083950042724609,
      "learning_rate": 9.966546331768191e-05,
      "loss": 4.6466,
      "step": 17
    },
    {
      "epoch": 0.17266187050359713,
      "grad_norm": 8.668869018554688,
      "learning_rate": 9.956320346634876e-05,
      "loss": 4.9438,
      "step": 18
    },
    {
      "epoch": 0.18225419664268586,
      "grad_norm": 8.577130317687988,
      "learning_rate": 9.944739353007344e-05,
      "loss": 4.927,
      "step": 19
    },
    {
      "epoch": 0.19184652278177458,
      "grad_norm": 9.07558536529541,
      "learning_rate": 9.931806517013612e-05,
      "loss": 4.7616,
      "step": 20
    },
    {
      "epoch": 0.2014388489208633,
      "grad_norm": 9.887092590332031,
      "learning_rate": 9.917525374361912e-05,
      "loss": 4.8108,
      "step": 21
    },
    {
      "epoch": 0.21103117505995203,
      "grad_norm": 8.207897186279297,
      "learning_rate": 9.901899829374047e-05,
      "loss": 4.8712,
      "step": 22
    },
    {
      "epoch": 0.22062350119904076,
      "grad_norm": 9.339859008789062,
      "learning_rate": 9.884934153917997e-05,
      "loss": 4.7193,
      "step": 23
    },
    {
      "epoch": 0.2302158273381295,
      "grad_norm": 10.292285919189453,
      "learning_rate": 9.86663298624003e-05,
      "loss": 4.8241,
      "step": 24
    },
    {
      "epoch": 0.23980815347721823,
      "grad_norm": 10.52551555633545,
      "learning_rate": 9.847001329696653e-05,
      "loss": 4.8344,
      "step": 25
    },
    {
      "epoch": 0.24940047961630696,
      "grad_norm": 11.95048999786377,
      "learning_rate": 9.826044551386744e-05,
      "loss": 4.2178,
      "step": 26
    },
    {
      "epoch": 0.2589928057553957,
      "grad_norm": 6.02551794052124,
      "learning_rate": 9.803768380684242e-05,
      "loss": 4.5288,
      "step": 27
    },
    {
      "epoch": 0.2685851318944844,
      "grad_norm": 4.607110977172852,
      "learning_rate": 9.780178907671789e-05,
      "loss": 4.087,
      "step": 28
    },
    {
      "epoch": 0.27817745803357313,
      "grad_norm": 3.7095556259155273,
      "learning_rate": 9.755282581475769e-05,
      "loss": 4.5215,
      "step": 29
    },
    {
      "epoch": 0.28776978417266186,
      "grad_norm": 3.812565803527832,
      "learning_rate": 9.729086208503174e-05,
      "loss": 4.5557,
      "step": 30
    },
    {
      "epoch": 0.2973621103117506,
      "grad_norm": 3.9195430278778076,
      "learning_rate": 9.701596950580806e-05,
      "loss": 4.5578,
      "step": 31
    },
    {
      "epoch": 0.3069544364508393,
      "grad_norm": 4.089346885681152,
      "learning_rate": 9.672822322997305e-05,
      "loss": 4.3776,
      "step": 32
    },
    {
      "epoch": 0.31654676258992803,
      "grad_norm": 3.834533929824829,
      "learning_rate": 9.642770192448536e-05,
      "loss": 4.3832,
      "step": 33
    },
    {
      "epoch": 0.3261390887290168,
      "grad_norm": 6.425683975219727,
      "learning_rate": 9.611448774886924e-05,
      "loss": 4.4965,
      "step": 34
    },
    {
      "epoch": 0.33573141486810554,
      "grad_norm": 5.3204026222229,
      "learning_rate": 9.578866633275288e-05,
      "loss": 4.5379,
      "step": 35
    },
    {
      "epoch": 0.34532374100719426,
      "grad_norm": 4.46568489074707,
      "learning_rate": 9.545032675245813e-05,
      "loss": 4.2879,
      "step": 36
    },
    {
      "epoch": 0.354916067146283,
      "grad_norm": 5.026806831359863,
      "learning_rate": 9.509956150664796e-05,
      "loss": 4.5137,
      "step": 37
    },
    {
      "epoch": 0.3645083932853717,
      "grad_norm": 5.191517353057861,
      "learning_rate": 9.473646649103818e-05,
      "loss": 4.6588,
      "step": 38
    },
    {
      "epoch": 0.37410071942446044,
      "grad_norm": 4.745547771453857,
      "learning_rate": 9.43611409721806e-05,
      "loss": 4.3923,
      "step": 39
    },
    {
      "epoch": 0.38369304556354916,
      "grad_norm": 5.328926086425781,
      "learning_rate": 9.397368756032445e-05,
      "loss": 4.4107,
      "step": 40
    },
    {
      "epoch": 0.3932853717026379,
      "grad_norm": 5.84630823135376,
      "learning_rate": 9.357421218136386e-05,
      "loss": 4.5784,
      "step": 41
    },
    {
      "epoch": 0.4028776978417266,
      "grad_norm": 5.517574310302734,
      "learning_rate": 9.316282404787871e-05,
      "loss": 4.3861,
      "step": 42
    },
    {
      "epoch": 0.41247002398081534,
      "grad_norm": 6.008972644805908,
      "learning_rate": 9.273963562927695e-05,
      "loss": 4.2943,
      "step": 43
    },
    {
      "epoch": 0.42206235011990406,
      "grad_norm": 6.8178277015686035,
      "learning_rate": 9.230476262104677e-05,
      "loss": 4.6216,
      "step": 44
    },
    {
      "epoch": 0.4316546762589928,
      "grad_norm": 7.200204372406006,
      "learning_rate": 9.185832391312644e-05,
      "loss": 4.7735,
      "step": 45
    },
    {
      "epoch": 0.4412470023980815,
      "grad_norm": 6.682967662811279,
      "learning_rate": 9.140044155740101e-05,
      "loss": 4.3549,
      "step": 46
    },
    {
      "epoch": 0.45083932853717024,
      "grad_norm": 7.509369373321533,
      "learning_rate": 9.093124073433463e-05,
      "loss": 4.7179,
      "step": 47
    },
    {
      "epoch": 0.460431654676259,
      "grad_norm": 7.858036518096924,
      "learning_rate": 9.045084971874738e-05,
      "loss": 4.7058,
      "step": 48
    },
    {
      "epoch": 0.47002398081534774,
      "grad_norm": 7.209802627563477,
      "learning_rate": 8.995939984474624e-05,
      "loss": 4.4991,
      "step": 49
    },
    {
      "epoch": 0.47961630695443647,
      "grad_norm": 12.480533599853516,
      "learning_rate": 8.945702546981969e-05,
      "loss": 4.4851,
      "step": 50
    },
    {
      "epoch": 0.47961630695443647,
      "eval_loss": 1.1107643842697144,
      "eval_runtime": 13.9632,
      "eval_samples_per_second": 12.605,
      "eval_steps_per_second": 3.151,
      "step": 50
    },
    {
      "epoch": 0.4892086330935252,
      "grad_norm": 9.453332901000977,
      "learning_rate": 8.894386393810563e-05,
      "loss": 4.3748,
      "step": 51
    },
    {
      "epoch": 0.4988009592326139,
      "grad_norm": 11.136406898498535,
      "learning_rate": 8.842005554284296e-05,
      "loss": 4.0162,
      "step": 52
    },
    {
      "epoch": 0.5083932853717026,
      "grad_norm": 8.412793159484863,
      "learning_rate": 8.788574348801675e-05,
      "loss": 4.3499,
      "step": 53
    },
    {
      "epoch": 0.5179856115107914,
      "grad_norm": 5.202061176300049,
      "learning_rate": 8.73410738492077e-05,
      "loss": 4.6036,
      "step": 54
    },
    {
      "epoch": 0.5275779376498801,
      "grad_norm": 4.200978755950928,
      "learning_rate": 8.678619553365659e-05,
      "loss": 4.3158,
      "step": 55
    },
    {
      "epoch": 0.5371702637889688,
      "grad_norm": 3.3434700965881348,
      "learning_rate": 8.622126023955446e-05,
      "loss": 4.3203,
      "step": 56
    },
    {
      "epoch": 0.5467625899280576,
      "grad_norm": 3.4814162254333496,
      "learning_rate": 8.564642241456986e-05,
      "loss": 4.0414,
      "step": 57
    },
    {
      "epoch": 0.5563549160671463,
      "grad_norm": 4.065919399261475,
      "learning_rate": 8.506183921362443e-05,
      "loss": 4.5392,
      "step": 58
    },
    {
      "epoch": 0.565947242206235,
      "grad_norm": 4.1824727058410645,
      "learning_rate": 8.44676704559283e-05,
      "loss": 4.3864,
      "step": 59
    },
    {
      "epoch": 0.5755395683453237,
      "grad_norm": 4.185287952423096,
      "learning_rate": 8.386407858128706e-05,
      "loss": 4.0991,
      "step": 60
    },
    {
      "epoch": 0.5851318944844125,
      "grad_norm": 5.290341377258301,
      "learning_rate": 8.32512286056924e-05,
      "loss": 4.5134,
      "step": 61
    },
    {
      "epoch": 0.5947242206235012,
      "grad_norm": 7.045202732086182,
      "learning_rate": 8.262928807620843e-05,
      "loss": 4.5054,
      "step": 62
    },
    {
      "epoch": 0.60431654676259,
      "grad_norm": 5.749582767486572,
      "learning_rate": 8.199842702516583e-05,
      "loss": 4.5051,
      "step": 63
    },
    {
      "epoch": 0.6139088729016786,
      "grad_norm": 5.410210609436035,
      "learning_rate": 8.135881792367686e-05,
      "loss": 4.4318,
      "step": 64
    },
    {
      "epoch": 0.6235011990407674,
      "grad_norm": 5.582022190093994,
      "learning_rate": 8.07106356344834e-05,
      "loss": 4.1789,
      "step": 65
    },
    {
      "epoch": 0.6330935251798561,
      "grad_norm": 6.269040584564209,
      "learning_rate": 8.005405736415126e-05,
      "loss": 4.4338,
      "step": 66
    },
    {
      "epoch": 0.6426858513189448,
      "grad_norm": 5.708898067474365,
      "learning_rate": 7.938926261462366e-05,
      "loss": 4.2831,
      "step": 67
    },
    {
      "epoch": 0.6522781774580336,
      "grad_norm": 6.113489627838135,
      "learning_rate": 7.871643313414718e-05,
      "loss": 4.1946,
      "step": 68
    },
    {
      "epoch": 0.6618705035971223,
      "grad_norm": 6.7462687492370605,
      "learning_rate": 7.803575286758364e-05,
      "loss": 4.5493,
      "step": 69
    },
    {
      "epoch": 0.6714628297362111,
      "grad_norm": 6.106730937957764,
      "learning_rate": 7.734740790612136e-05,
      "loss": 4.2402,
      "step": 70
    },
    {
      "epoch": 0.6810551558752997,
      "grad_norm": 6.1799163818359375,
      "learning_rate": 7.66515864363997e-05,
      "loss": 4.3191,
      "step": 71
    },
    {
      "epoch": 0.6906474820143885,
      "grad_norm": 6.90517520904541,
      "learning_rate": 7.594847868906076e-05,
      "loss": 4.6124,
      "step": 72
    },
    {
      "epoch": 0.7002398081534772,
      "grad_norm": 8.273045539855957,
      "learning_rate": 7.52382768867422e-05,
      "loss": 4.1286,
      "step": 73
    },
    {
      "epoch": 0.709832134292566,
      "grad_norm": 7.589626312255859,
      "learning_rate": 7.452117519152542e-05,
      "loss": 4.4932,
      "step": 74
    },
    {
      "epoch": 0.7194244604316546,
      "grad_norm": 7.502345085144043,
      "learning_rate": 7.379736965185368e-05,
      "loss": 4.0907,
      "step": 75
    },
    {
      "epoch": 0.7290167865707434,
      "grad_norm": 8.074150085449219,
      "learning_rate": 7.30670581489344e-05,
      "loss": 4.1704,
      "step": 76
    },
    {
      "epoch": 0.7386091127098321,
      "grad_norm": 9.586164474487305,
      "learning_rate": 7.233044034264034e-05,
      "loss": 4.4583,
      "step": 77
    },
    {
      "epoch": 0.7482014388489209,
      "grad_norm": 10.162365913391113,
      "learning_rate": 7.158771761692464e-05,
      "loss": 3.6437,
      "step": 78
    },
    {
      "epoch": 0.7577937649880095,
      "grad_norm": 7.662997722625732,
      "learning_rate": 7.083909302476453e-05,
      "loss": 4.2468,
      "step": 79
    },
    {
      "epoch": 0.7673860911270983,
      "grad_norm": 5.551909446716309,
      "learning_rate": 7.008477123264848e-05,
      "loss": 4.6284,
      "step": 80
    },
    {
      "epoch": 0.7769784172661871,
      "grad_norm": 3.8176558017730713,
      "learning_rate": 6.932495846462261e-05,
      "loss": 4.0981,
      "step": 81
    },
    {
      "epoch": 0.7865707434052758,
      "grad_norm": 3.187798023223877,
      "learning_rate": 6.855986244591104e-05,
      "loss": 4.0516,
      "step": 82
    },
    {
      "epoch": 0.7961630695443646,
      "grad_norm": 3.0702691078186035,
      "learning_rate": 6.778969234612584e-05,
      "loss": 4.0149,
      "step": 83
    },
    {
      "epoch": 0.8057553956834532,
      "grad_norm": 3.2330820560455322,
      "learning_rate": 6.701465872208216e-05,
      "loss": 4.031,
      "step": 84
    },
    {
      "epoch": 0.815347721822542,
      "grad_norm": 3.4017140865325928,
      "learning_rate": 6.623497346023418e-05,
      "loss": 4.2663,
      "step": 85
    },
    {
      "epoch": 0.8249400479616307,
      "grad_norm": 3.7491283416748047,
      "learning_rate": 6.545084971874738e-05,
      "loss": 4.2055,
      "step": 86
    },
    {
      "epoch": 0.8345323741007195,
      "grad_norm": 3.9540011882781982,
      "learning_rate": 6.466250186922325e-05,
      "loss": 4.2015,
      "step": 87
    },
    {
      "epoch": 0.8441247002398081,
      "grad_norm": 4.023013591766357,
      "learning_rate": 6.387014543809223e-05,
      "loss": 4.2716,
      "step": 88
    },
    {
      "epoch": 0.8537170263788969,
      "grad_norm": 5.721251487731934,
      "learning_rate": 6.307399704769099e-05,
      "loss": 4.3664,
      "step": 89
    },
    {
      "epoch": 0.8633093525179856,
      "grad_norm": 6.0954999923706055,
      "learning_rate": 6.227427435703997e-05,
      "loss": 4.096,
      "step": 90
    },
    {
      "epoch": 0.8729016786570744,
      "grad_norm": 5.155120372772217,
      "learning_rate": 6.147119600233758e-05,
      "loss": 4.1793,
      "step": 91
    },
    {
      "epoch": 0.882494004796163,
      "grad_norm": 5.015108108520508,
      "learning_rate": 6.066498153718735e-05,
      "loss": 4.2802,
      "step": 92
    },
    {
      "epoch": 0.8920863309352518,
      "grad_norm": 5.613718509674072,
      "learning_rate": 5.985585137257401e-05,
      "loss": 4.338,
      "step": 93
    },
    {
      "epoch": 0.9016786570743405,
      "grad_norm": 6.009364604949951,
      "learning_rate": 5.90440267166055e-05,
      "loss": 3.948,
      "step": 94
    },
    {
      "epoch": 0.9112709832134293,
      "grad_norm": 5.694631099700928,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 4.1477,
      "step": 95
    },
    {
      "epoch": 0.920863309352518,
      "grad_norm": 5.532630443572998,
      "learning_rate": 5.74131823855921e-05,
      "loss": 4.1783,
      "step": 96
    },
    {
      "epoch": 0.9304556354916067,
      "grad_norm": 6.125311851501465,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 4.149,
      "step": 97
    },
    {
      "epoch": 0.9400479616306955,
      "grad_norm": 6.07303524017334,
      "learning_rate": 5.577423184847932e-05,
      "loss": 3.9012,
      "step": 98
    },
    {
      "epoch": 0.9496402877697842,
      "grad_norm": 6.486006736755371,
      "learning_rate": 5.495227651252315e-05,
      "loss": 4.0906,
      "step": 99
    },
    {
      "epoch": 0.9592326139088729,
      "grad_norm": 7.411128997802734,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 4.2295,
      "step": 100
    },
    {
      "epoch": 0.9592326139088729,
      "eval_loss": 1.0107861757278442,
      "eval_runtime": 14.245,
      "eval_samples_per_second": 12.355,
      "eval_steps_per_second": 3.089,
      "step": 100
    },
    {
      "epoch": 0.9688249400479616,
      "grad_norm": 7.739264011383057,
      "learning_rate": 5.330452921628497e-05,
      "loss": 4.2527,
      "step": 101
    },
    {
      "epoch": 0.9784172661870504,
      "grad_norm": 7.676518440246582,
      "learning_rate": 5.247918773366112e-05,
      "loss": 3.966,
      "step": 102
    },
    {
      "epoch": 0.988009592326139,
      "grad_norm": 8.727714538574219,
      "learning_rate": 5.165316846586541e-05,
      "loss": 3.9285,
      "step": 103
    },
    {
      "epoch": 0.9976019184652278,
      "grad_norm": 11.266743659973145,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 3.8618,
      "step": 104
    },
    {
      "epoch": 1.0071942446043165,
      "grad_norm": 4.511717319488525,
      "learning_rate": 5e-05,
      "loss": 3.736,
      "step": 105
    },
    {
      "epoch": 1.0167865707434052,
      "grad_norm": 4.291499614715576,
      "learning_rate": 4.917330276168208e-05,
      "loss": 3.6386,
      "step": 106
    },
    {
      "epoch": 1.026378896882494,
      "grad_norm": 5.089818000793457,
      "learning_rate": 4.834683153413459e-05,
      "loss": 3.6019,
      "step": 107
    },
    {
      "epoch": 1.0359712230215827,
      "grad_norm": 3.8521947860717773,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 3.403,
      "step": 108
    },
    {
      "epoch": 1.0455635491606714,
      "grad_norm": 3.1861042976379395,
      "learning_rate": 4.669547078371504e-05,
      "loss": 3.3043,
      "step": 109
    },
    {
      "epoch": 1.0551558752997603,
      "grad_norm": 3.1751973628997803,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 3.1966,
      "step": 110
    },
    {
      "epoch": 1.064748201438849,
      "grad_norm": 3.4896082878112793,
      "learning_rate": 4.504772348747687e-05,
      "loss": 3.1622,
      "step": 111
    },
    {
      "epoch": 1.0743405275779376,
      "grad_norm": 4.223556041717529,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 3.3034,
      "step": 112
    },
    {
      "epoch": 1.0839328537170263,
      "grad_norm": 4.142705917358398,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 3.0643,
      "step": 113
    },
    {
      "epoch": 1.0935251798561152,
      "grad_norm": 4.401920795440674,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 3.1777,
      "step": 114
    },
    {
      "epoch": 1.1031175059952039,
      "grad_norm": 4.815911769866943,
      "learning_rate": 4.17702704859633e-05,
      "loss": 3.0945,
      "step": 115
    },
    {
      "epoch": 1.1127098321342925,
      "grad_norm": 4.681511402130127,
      "learning_rate": 4.095597328339452e-05,
      "loss": 3.0561,
      "step": 116
    },
    {
      "epoch": 1.1223021582733812,
      "grad_norm": 4.726698398590088,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.995,
      "step": 117
    },
    {
      "epoch": 1.13189448441247,
      "grad_norm": 5.100158214569092,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.9533,
      "step": 118
    },
    {
      "epoch": 1.1414868105515588,
      "grad_norm": 5.334418773651123,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.8864,
      "step": 119
    },
    {
      "epoch": 1.1510791366906474,
      "grad_norm": 5.059926986694336,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.5449,
      "step": 120
    },
    {
      "epoch": 1.160671462829736,
      "grad_norm": 5.714264869689941,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.6519,
      "step": 121
    },
    {
      "epoch": 1.170263788968825,
      "grad_norm": 6.931163787841797,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.7539,
      "step": 122
    },
    {
      "epoch": 1.1798561151079137,
      "grad_norm": 7.079442501068115,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.7795,
      "step": 123
    },
    {
      "epoch": 1.1894484412470023,
      "grad_norm": 6.2183837890625,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.4587,
      "step": 124
    },
    {
      "epoch": 1.1990407673860912,
      "grad_norm": 6.992701053619385,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.384,
      "step": 125
    },
    {
      "epoch": 1.20863309352518,
      "grad_norm": 7.548563480377197,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.6901,
      "step": 126
    },
    {
      "epoch": 1.2182254196642686,
      "grad_norm": 8.789972305297852,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.6052,
      "step": 127
    },
    {
      "epoch": 1.2278177458033572,
      "grad_norm": 8.276802062988281,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.5424,
      "step": 128
    },
    {
      "epoch": 1.2374100719424461,
      "grad_norm": 9.502008438110352,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.0581,
      "step": 129
    },
    {
      "epoch": 1.2470023980815348,
      "grad_norm": 10.5198974609375,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.8548,
      "step": 130
    },
    {
      "epoch": 1.2565947242206235,
      "grad_norm": 5.544888496398926,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.9438,
      "step": 131
    },
    {
      "epoch": 1.2661870503597124,
      "grad_norm": 6.117003917694092,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 4.1228,
      "step": 132
    },
    {
      "epoch": 1.275779376498801,
      "grad_norm": 4.789488792419434,
      "learning_rate": 2.766955965735968e-05,
      "loss": 3.4874,
      "step": 133
    },
    {
      "epoch": 1.2853717026378897,
      "grad_norm": 5.141814708709717,
      "learning_rate": 2.693294185106562e-05,
      "loss": 3.4811,
      "step": 134
    },
    {
      "epoch": 1.2949640287769784,
      "grad_norm": 4.24757719039917,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 3.2962,
      "step": 135
    },
    {
      "epoch": 1.304556354916067,
      "grad_norm": 4.041790008544922,
      "learning_rate": 2.547882480847461e-05,
      "loss": 3.0472,
      "step": 136
    },
    {
      "epoch": 1.314148681055156,
      "grad_norm": 3.7918574810028076,
      "learning_rate": 2.476172311325783e-05,
      "loss": 3.1758,
      "step": 137
    },
    {
      "epoch": 1.3237410071942446,
      "grad_norm": 3.4599804878234863,
      "learning_rate": 2.405152131093926e-05,
      "loss": 3.2114,
      "step": 138
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 3.5690040588378906,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.8981,
      "step": 139
    },
    {
      "epoch": 1.3429256594724222,
      "grad_norm": 3.8757100105285645,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 3.0871,
      "step": 140
    },
    {
      "epoch": 1.3525179856115108,
      "grad_norm": 3.9097936153411865,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.992,
      "step": 141
    },
    {
      "epoch": 1.3621103117505995,
      "grad_norm": 4.1639509201049805,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.9172,
      "step": 142
    },
    {
      "epoch": 1.3717026378896882,
      "grad_norm": 4.370047092437744,
      "learning_rate": 2.061073738537635e-05,
      "loss": 3.1956,
      "step": 143
    },
    {
      "epoch": 1.381294964028777,
      "grad_norm": 4.603968143463135,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.7496,
      "step": 144
    },
    {
      "epoch": 1.3908872901678657,
      "grad_norm": 4.59792423248291,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.7783,
      "step": 145
    },
    {
      "epoch": 1.4004796163069544,
      "grad_norm": 5.105554580688477,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.7469,
      "step": 146
    },
    {
      "epoch": 1.4100719424460433,
      "grad_norm": 5.359896183013916,
      "learning_rate": 1.800157297483417e-05,
      "loss": 3.0056,
      "step": 147
    },
    {
      "epoch": 1.419664268585132,
      "grad_norm": 5.563641548156738,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.8243,
      "step": 148
    },
    {
      "epoch": 1.4292565947242206,
      "grad_norm": 6.391464710235596,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.7181,
      "step": 149
    },
    {
      "epoch": 1.4388489208633093,
      "grad_norm": 6.547914028167725,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.7078,
      "step": 150
    },
    {
      "epoch": 1.4388489208633093,
      "eval_loss": 1.0176074504852295,
      "eval_runtime": 14.231,
      "eval_samples_per_second": 12.367,
      "eval_steps_per_second": 3.092,
      "step": 150
    },
    {
      "epoch": 1.448441247002398,
      "grad_norm": 6.124886512756348,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.4631,
      "step": 151
    },
    {
      "epoch": 1.4580335731414868,
      "grad_norm": 7.960712432861328,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.8563,
      "step": 152
    },
    {
      "epoch": 1.4676258992805755,
      "grad_norm": 8.012553215026855,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.7041,
      "step": 153
    },
    {
      "epoch": 1.4772182254196642,
      "grad_norm": 7.600780010223389,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.4972,
      "step": 154
    },
    {
      "epoch": 1.486810551558753,
      "grad_norm": 9.274397850036621,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.5683,
      "step": 155
    },
    {
      "epoch": 1.4964028776978417,
      "grad_norm": 10.070465087890625,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.0744,
      "step": 156
    },
    {
      "epoch": 1.5059952038369304,
      "grad_norm": 5.811046123504639,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.7763,
      "step": 157
    },
    {
      "epoch": 1.515587529976019,
      "grad_norm": 4.000051975250244,
      "learning_rate": 1.157994445715706e-05,
      "loss": 3.7692,
      "step": 158
    },
    {
      "epoch": 1.5251798561151078,
      "grad_norm": 4.379804611206055,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 3.5487,
      "step": 159
    },
    {
      "epoch": 1.5347721822541966,
      "grad_norm": 3.8836669921875,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 3.2801,
      "step": 160
    },
    {
      "epoch": 1.5443645083932853,
      "grad_norm": 4.066531181335449,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 3.0887,
      "step": 161
    },
    {
      "epoch": 1.5539568345323742,
      "grad_norm": 4.297153472900391,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.9448,
      "step": 162
    },
    {
      "epoch": 1.5635491606714629,
      "grad_norm": 4.353287220001221,
      "learning_rate": 9.068759265665384e-06,
      "loss": 3.1022,
      "step": 163
    },
    {
      "epoch": 1.5731414868105515,
      "grad_norm": 5.311435222625732,
      "learning_rate": 8.599558442598998e-06,
      "loss": 3.2913,
      "step": 164
    },
    {
      "epoch": 1.5827338129496402,
      "grad_norm": 4.701266765594482,
      "learning_rate": 8.141676086873572e-06,
      "loss": 3.1336,
      "step": 165
    },
    {
      "epoch": 1.5923261390887289,
      "grad_norm": 4.6634979248046875,
      "learning_rate": 7.695237378953223e-06,
      "loss": 3.0941,
      "step": 166
    },
    {
      "epoch": 1.6019184652278178,
      "grad_norm": 4.581821918487549,
      "learning_rate": 7.260364370723044e-06,
      "loss": 3.1542,
      "step": 167
    },
    {
      "epoch": 1.6115107913669064,
      "grad_norm": 4.738865375518799,
      "learning_rate": 6.837175952121306e-06,
      "loss": 3.0948,
      "step": 168
    },
    {
      "epoch": 1.6211031175059953,
      "grad_norm": 4.442863464355469,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.6232,
      "step": 169
    },
    {
      "epoch": 1.630695443645084,
      "grad_norm": 4.823970794677734,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.9542,
      "step": 170
    },
    {
      "epoch": 1.6402877697841727,
      "grad_norm": 5.096918106079102,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.8724,
      "step": 171
    },
    {
      "epoch": 1.6498800959232613,
      "grad_norm": 4.839364528656006,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.6571,
      "step": 172
    },
    {
      "epoch": 1.65947242206235,
      "grad_norm": 4.870880126953125,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.3409,
      "step": 173
    },
    {
      "epoch": 1.6690647482014387,
      "grad_norm": 5.062325954437256,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.3736,
      "step": 174
    },
    {
      "epoch": 1.6786570743405276,
      "grad_norm": 5.490885257720947,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.6341,
      "step": 175
    },
    {
      "epoch": 1.6882494004796165,
      "grad_norm": 5.871090888977051,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.6431,
      "step": 176
    },
    {
      "epoch": 1.6978417266187051,
      "grad_norm": 6.174441814422607,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.4706,
      "step": 177
    },
    {
      "epoch": 1.7074340527577938,
      "grad_norm": 6.304491996765137,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.5512,
      "step": 178
    },
    {
      "epoch": 1.7170263788968825,
      "grad_norm": 7.1342010498046875,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.5674,
      "step": 179
    },
    {
      "epoch": 1.7266187050359711,
      "grad_norm": 7.464678764343262,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.4473,
      "step": 180
    },
    {
      "epoch": 1.7362110311750598,
      "grad_norm": 8.325981140136719,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.5243,
      "step": 181
    },
    {
      "epoch": 1.7458033573141487,
      "grad_norm": 10.566821098327637,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.6227,
      "step": 182
    },
    {
      "epoch": 1.7553956834532374,
      "grad_norm": 3.730330228805542,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.8606,
      "step": 183
    },
    {
      "epoch": 1.7649880095923263,
      "grad_norm": 2.7509052753448486,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 3.7177,
      "step": 184
    },
    {
      "epoch": 1.774580335731415,
      "grad_norm": 2.8649446964263916,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 3.275,
      "step": 185
    },
    {
      "epoch": 1.7841726618705036,
      "grad_norm": 3.036092519760132,
      "learning_rate": 1.333670137599713e-06,
      "loss": 3.2502,
      "step": 186
    },
    {
      "epoch": 1.7937649880095923,
      "grad_norm": 3.3191261291503906,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 3.1055,
      "step": 187
    },
    {
      "epoch": 1.803357314148681,
      "grad_norm": 3.3392696380615234,
      "learning_rate": 9.810017062595322e-07,
      "loss": 3.1448,
      "step": 188
    },
    {
      "epoch": 1.8129496402877698,
      "grad_norm": 3.43424391746521,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.9319,
      "step": 189
    },
    {
      "epoch": 1.8225419664268585,
      "grad_norm": 3.7430198192596436,
      "learning_rate": 6.819348298638839e-07,
      "loss": 3.1315,
      "step": 190
    },
    {
      "epoch": 1.8321342925659474,
      "grad_norm": 4.130662441253662,
      "learning_rate": 5.526064699265753e-07,
      "loss": 3.1249,
      "step": 191
    },
    {
      "epoch": 1.841726618705036,
      "grad_norm": 4.219359874725342,
      "learning_rate": 4.367965336512403e-07,
      "loss": 3.0319,
      "step": 192
    },
    {
      "epoch": 1.8513189448441247,
      "grad_norm": 4.476027488708496,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.9273,
      "step": 193
    },
    {
      "epoch": 1.8609112709832134,
      "grad_norm": 4.654909133911133,
      "learning_rate": 2.458548727494292e-07,
      "loss": 3.0786,
      "step": 194
    },
    {
      "epoch": 1.870503597122302,
      "grad_norm": 4.479025363922119,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.7494,
      "step": 195
    },
    {
      "epoch": 1.8800959232613907,
      "grad_norm": 5.146566867828369,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.6221,
      "step": 196
    },
    {
      "epoch": 1.8896882494004796,
      "grad_norm": 4.966414451599121,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.7186,
      "step": 197
    },
    {
      "epoch": 1.8992805755395683,
      "grad_norm": 4.95269775390625,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.4959,
      "step": 198
    },
    {
      "epoch": 1.9088729016786572,
      "grad_norm": 5.35521125793457,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.8723,
      "step": 199
    },
    {
      "epoch": 1.9184652278177459,
      "grad_norm": 5.235119342803955,
      "learning_rate": 0.0,
      "loss": 2.5195,
      "step": 200
    },
    {
      "epoch": 1.9184652278177459,
      "eval_loss": 0.9910973906517029,
      "eval_runtime": 14.2459,
      "eval_samples_per_second": 12.354,
      "eval_steps_per_second": 3.089,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.974243271863173e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}